ROCView

What is ROCView?

ROCView [1] has been developed to provide an image display and response capture (IDRC) solution to free-response receiver operating characteristic (FROC) analysis. The web-based software allows the collection of data for multiple localisations and confidence scores (mark-rating pairs) for each image (case). The software allows an administrator to set the ROC truth, to which all subsequent observations are compared. A proximity criterion is used to classify observer localisations as either a lesion localisation (LL) or non-lesion localisation (NL), separating marks or lesions and lesion-mimics. Localisations are then saved in either a true positive or false positive worksheet that can be downloaded. Each worksheet includes the observer ID, case ID lesion ID and TP/FP rating; these *.csv files can be easily adapted for jackknife alternative free-response receiver operating characteristic (JAFROC) analysis [2].

ROCView allows comparison of multiple modalities by multiple observers and can run concurrent studies. Proximity criterion (radius) can be changed for each study. Several studies have successfully used ROCView to date [3,4,5], with others in progress.

ROCView was designed by John Thompson and developed by Stephen Thompson..

1. Thompson, J. Hogg, P. Thompson, S. Manning, D. Szczepura, K. ROCView: Prototype software for data collection in jackknife alternative free-response receiver operating characteristic analysis. The British Journal of Radiology 2012:doi:10.1259/bjr/99497945
2. Chakraborty, D. The ROC Task. http://www.devchakraborty.com/Receiver%20operating%20characteristic.pdf
3. Thompson, J. Hogg, P. Szczepura, K. Manning, D. Analysis of CT acquisition parameters suitable for use in SPECT/CT: A free-response receiver operating characteristic study. Radiography 2012:doi:10.1016/j.radi.2012.05.004
4. Thompson, J. Szczepura, K. Manning, D. Hogg, P. Lesion detection in the CT attenuation correction image of 5 different low resolution SPECT/CT systems: a multi-centre study. Proceedings of the UK Radiological Congress (UKRC) 2012
5. Thompson, J. Higham, S. Hogg, P. Manning, D. Szczepura, K. The impact of tube current variation on lesion detection in the attenuation correction image co-incidentally acquired for myocardial perfusion imaging in SPECT/CT: a phantom based study. Proceedings of the UK Radiological Congress (UKRC) 2012

Use

ROCView evaluations are available at do.rocview.net

Contact